Bug #8450 for POE: patch for argument handling by POE::Session subclasses

You explained that it's cleaner and faster to go with your patch, but I'm not convinced. The patch adds a bit of overhead to all session constructions, and it seems to perform the moral equivalent of sub create { my ($class, %params) = @_; # Validate and extract the subclass parameters. # Something like this, but maybe more optimal. my $param = delete $params{param} || croak "hey, you need a param"; # Create the session. my $self = bless $class->SUPER::create(@params), $class; # Plug the values into $self. $self->[WOOT] = $subclass_params{param}; # Return the object. Not that we care. return $self; }

We discussed this in IRC. Martijn said it was okay to close the ticket without applying the patch.

When I agreed the proposed solution was ok, I hadn't looked into this for a long time. Turns out it doesn't actually work for what I want, because I am overriding the _start event. I do this in try_alloc, which gets called during create because then the original _start event (if any) is available in a nicely managable form. So the reason for the patch which started this ticket is that I need these custom arguments at try_alloc() time, so before $class->SUPER::create is finished, and I couldn't store them in $self before $class->SUPER::create is finished